Event overview

Competition concept

　As a new technological field called CASE becomes the center of competition in the automobile industry, there is a shortage of AI and IT engineers, and the need for them is rapidly increasing throughout the automobile industry. 　As a new initiative to discover and develop engineers who will lead the automotive industry in the future, we will hold an international competition to compete in AI technology for autonomous driving. 　Using this competition, which is held in cooperation between industry, government, and academia, as a starting point, we aim to build a framework that will contribute to the further development of the automobile industry.

Purpose of Japan Automotive AI Challenge 2023 (Simulation)

Motorsports, which test the driving performance of automobiles under extreme conditions, have contributed to the improvement and development of automobile performance by creating technical challenges. Japan Automotive AI Challenge 2023 (Simulation) aims to contribute to the development of autonomous driving technology that can drive safely and at high speeds, with the theme of racing in an End to End (E2E) simulation environment.

Automated driving in races involves a variety of technological elements, including control technology that takes into account skidding, air resistance, and vehicle response characteristics, as well as tracking and behavior prediction of other vehicles. This conference will provide participants with an opportunity to incorporate these elements into developing autonomous driving software with strategic driving planning and advanced vehicle control functions, and to share their knowledge. We hope that the engineers who acquire new skills through this challenge will contribute to technological innovation in the automobile industry.

3 challenges

Challenge 1: Robust Perception

Self-driving cars require a high level of safety. The first step to avoiding a collision is to recognize other cars around you 360 degrees from sensor data and predict their movements. The first challenge is to develop a fast and robust recognition function.

Challenge 2: Strategic Route Planning

The second challenge is strategic path planning. In order to travel faster than other vehicles, it is necessary to use recognition results to process the ever-changing surrounding environment and constantly seek the optimal route.

Challenge 3: Control in High-Speed Driving

Control technology that realizes optimal route planning is essential to winning races. The third challenge is to develop a control module that can operate at high speeds, allowing the vehicle to travel along strategic and complex routes.

Competition Description - Racing

Overview

Participants will develop autonomous driving software based on Autoware.Universeand integrate it into a racing car running in the End to End Simulation Space (AWSIM). The goal is to win the race while driving safely using the developed self-driving software.

Driving course and rivals

The running course for this event is approximately 5.8km long and has a complex shape, including long straights and hairpin curves. You will drive along this circuit-themed course with multiple rival cars. Rival cars will accelerate, decelerate, and change course to match your self-driving car, requiring strategic route planning.

About evaluation criteria

The goal of this tournament is to reach the goal quickly and safely. Therefore, participants' driving using autonomous driving software will be evaluated from the following three perspectives. Further details on scoring and rules will be made available in the participant documentation that will be published at a later date.

1. Mileage
2. Safety
3. Speed

About the development environment

• Participants must prepare a computer running Ubuntu 22.04as a development environment.
• Please submit your developed repository and we will grade it in a cloud environment.
• Details of the competition repository and competition content are scheduled to be released in early November.